Professional Documents
Culture Documents
Introduction To The Principles of Vacuum Physics: Niels Marquardt
Introduction To The Principles of Vacuum Physics: Niels Marquardt
Niels Marquardt
Institute for Accelerator Physics and Synchrotron Radiation,
University of Dortmund, 44221 Dortmund, Germany
Abstract
Vacuum physics is the necessary condition for scientific research and
modern high technology. In this introduction to the physics and technology
of vacuum the basic concepts of a gas composed of atoms and molecules are
presented. These gas particles are contained in a partially empty volume
forming the vacuum. The fundamentals of vacuum, molecular density,
pressure, velocity distribution, mean free path, particle velocity,
conductivity, temperature and gas flow are discussed.
that there is no friction in space, but vacuum instead. On the occasion of the Reichstag at
Regensburg/Germany he performed publicly in 1654 his famous and most spectacular experiment with
the two empty, gasketed copper "Magdeburg Hemispheres" of about 50 cm in diameter (see Fig. 4).
By this experiment he demonstrated that it is impossible to pull the two halves apart against the air
pressure, even by using 2 8 horses (the counter-pressure by air in the interior of the sphere is
missing). During this time, it became clear that we are living on the bottom of a huge ocean of air and
that the mass of the atmosphere corresponds to a pressure of about 1 kg per cm2 or 10 tons on an area
of 1 m2. This is the reason, why the 16 horses of von Guericke were unable to pull the hemispheres
apart. The reason, why we dont feel anything of this tremendous pressure is simply that there is the
same pressure inside our body. To determine the mass of air, Otto von Guericke performed his
spectacular gallows experiment, shown in Fig. 5.
Fig. 1 Experiments of Torricelli (1608-1647), observing the height of a mercury column in one-sided glass tubes to be
independent of the shape and inclination of the tubes. In the volume above the mercury meniscus was the first experimentally
generated vacuum, the "Torricelli Vacuum" (taken from an engraving of P. Gasparis Schotti "Technica Curiosa", Nrnberg,
1664. Courtesy C. Edelmann [11]).
3
Fig. 3 Portrait of Otto von Guericke (1602-1686), (Courtesy "Otto von Guericke Society" in Magdeburg).
4
Fig. 4 Experiment with the "Magdeburger Hemispheres" (1654) by Otto von Guericke (taken from an engraving of
P. Gasparis Schotti "Technica Curiosa", Nrnberg, 1664. Courtesy Otto von Guericke Society in Magdeburg).
Another important discovery was made by Robert Boyle (1627-1691, who built an improved
vacuum pump, and Edme Mariotte (~1620-1684). They derived the law of Boyle-Mariotte, the
fundamental equation of gas laws, valid for so-called "ideal gases" being in thermodynamical
equilibrium :
p V / T = constant (with p = pressure, V = volume and T = absolute temperature of the gas).
Many technical inventions, concerning various pumps for vacuum generation, were made in the
last, and until the middle, of our century, leading for instance to the discovery of the electron in 1897
by J.J. Thomson (1856-1940) and the invention of the X-ray tube by Wilhelm Conrad Rntgen (1845-
1923) with his discovery of short-wavelength X-ray radiation on November 8, 1895. Finally, in the
thirties of our century, the invention of the wide variety of particle accelerators started and went on till
today with the construction of huge machines. With accelerators, nuclear and high-energy elementary-
particle physics began. All the inventions and sophisticated technological developments of today
would have been unthinkable without high vacuum and its technology. Therefore, vacuum physics and
techniques paved the way for the development of modern high-technology and our industrial society.
What characterises vacuum? The particles in a volume are in constant motion. They hit the
walls of the container and exert a force on its surface area, which is called "pressure". One can
measure this force per unit area of the vessel, called pressure, by comparing it to the atmospheric
pressure or by determining it in absolute units. Therefore, by a pressure measurement one obtains the
number and intensity of particle impacts on a unit of surface area.
The highest pressures ever obtained in a laboratory happened in the European Laboratory for
High Energy Nuclear Physics CERN during head-on collisions of two fast lead nuclei. This is the
unimaginably large pressure of 1030 bar, which is estimated to be about 5-times larger than before a
supernova explosion. Compared with this value, the maximum static pressure of about 3.5 106 bar,
reached in experiments with diamond stamps, is very small. On the other hand, the lowest pressure or
5
Fig. 5 Gallows experiment of Otto von Guericke (1602-1686), in order to determine the mass of air (taken from an engraving
of Ottonis de Guericke "Experimenta nova (ut vocantur) Magdeburgica de vacuo spatio". Courtesy C. Edelmann [11]). In a
copper cylinder a , fixed at the gallows post on the right side of the picture, a piston Q can move vertically. The piston is
connected with a rope running over two rolls H and T to a board loaded with several weights. At first, the piston is pulled up
with the tap X at the bottom of the cylinder open. When evacuating the cylinder with an air pump, tap closed, the piston
slides downwards due to the outside air pressure, lifting the weights.
highest vacuum generated by pumps in the laboratory amounts to 10-17 bar, which corresponds to just a
few hundred particles in one cm3 and which was also obtained at CERN. One should compare this with
a number of ~ 3 1019 particles per cm3 at atmospheric pressure and room temperature. Therefore,
between the highest and lowest pressure there are 47 orders of magnitude in between. But even the
best vacuum obtained on earth is a high-pressure area compared with the almost total emptiness
between stars in space. Accordingly, besides pressure vacuum is characterised by the density of
particles. The interstellar particle density in the Milky Way, for instance, consisting of gas, plasma and
dust, is only ~ 5 104 particles per m3. Between Galaxies one has only one particle or at most a few of
them per m3. If one would distribute homogeneously the total matter of the universe in space, one
would still have an extremely low particle density of only 3 particles per m3. The universal density
scale, presented in Table 1, exhibits the enormous range of about 60 orders of magnitude between the
maximum density calculated for a black hole and almost totally empty regions in space with the earth
somewhere in the middle. Table 2 shows examples of particle densities of various objects. However,
even the almost perfect vacuum of interstellar space is not empty at all. There is electromagnetic
radiation everywhere or superstrong fields, which are even able to generate new particles. Even if we
would succeed to construct an empty volume, totally shielded against radiation from outside and
cooled to absolute zero temperature to reduce radiation from its walls, there would still be radiation
emitted from the particles of the walls which are never totally at rest (due to Heisenbergs uncertainty
principle).
6
Table 1
Universal density scale
3
Universal density scale g/cm
Big Bang
Black hole 10 27
Neutron Star 10 14
White Dwarf 10 6
Solar Center 158
Sun (average density) 1.4 ~10 6 0
Water on Earth 1
Air 10 -3
Vacuum in Laboratory 10 -18
Interstellar Space 10 -24
Universe 10 -30
Empty Regions in Space 10 -33
Table 2
Examples of particle densities of various objects.
Fig. 6 Arrangement of vacuum components in a typical UHV-system for surface studies [9].
Vacuum is made by removing the various gases from the contained volume. This is done by
many different kinds of vacuum pumps, which either remove the particles from the volume by real
pumping or trap them by binding them via physical or chemical forces, changing their form and
keeping them trapped in the bulk of the pumping material, a so-called getter. This is done by various
mechanical and sorption pumps, diffusion pumps, cryo-pumps, ion-sputter pumps, non-evaporable-
getter(NEG) pumps or titanium sublimation pumps. As an example for the tremendous development of
mechanical pumps during the second half of the 20th century, in particular of fast rotating turbo-
molecular pumps, Fig. 7 shows the so-called dry (oilfree) wide-range type of such a pump, where the
rotor is kept in position by magnetic bearings. Not only pumps for vacuum generation, but also a wide
variety of devices for pressure measurements, rest-gas analysis and mass spectrometry have been
developed for the various ranges of vacuum. All these highly sophisticated devices for vacuum
generation and measurement and the corresponding materials and cleaning techniques form the basis
for modern vacuum technology and will be discussed in detail in the following chapters.
8
There are many people who believe that vacuum technology is just dealing with valves, flanges
and other vacuum components. But the science and technology that all this hardware enables are the
keys to our high-tech society. Vacuum is used for a huge number of technical processes and products,
like vacuum-packed coffee, thermos flasks, light bulbs, microwave and cathode-ray tubes for TV
screens. One uses it always when one needs a "clean" space, which is free of gases and contaminants.
It led to the development of ion-getter pumps and ultra-high-vacuum (UHV) techniques, to electron
and ion microscopes, ion implanters, sensitive leak detectors and space simulation technology.
Because its conductivity is lower than any other material, vacuum is best suited for thermal insulation,
i.e. to reduce energy transfer or to prevent heat loss. It is used for thermal insulation at low
temperatures (below 100 K), in particular. On the other hand, low temperatures are used to generate
ultra-high vacuum by cryo-pumping. Contamination by gas particles can disturb or totally prevent
processes or spoil products. Therefore, vacuum technology and equipment was essential to the
development of electron and microwave tubes, the production of high-purity metals, the separation
and storage of gases, any plant producing computer chips and semi-conductor components, displays,
thin films, nanostructures, optical and microelectronic coatings, vacuum metallurgy, data storage and
hard disk drives. Vacuum technology is growing in importance as more industries introduce advanced
materials and devices and as technology goes more and more into the direction of components of
micro- and nanometer scale.
Ultra-high vacuum is also of particular importance for chemical and biological techniques, for
atomic-physics research and for the operation of modern particle accelerators, used for high-energy
elementary-particle physics or synchrotron-radiation research. The development of the so-called 3rd
generation synchrotron light sources with their tremendous fluxes of photons of all wavelength, used
nowadays in many areas of natural science, would have been impossible without UHV-technology. A
comparison of the brightness, i.e. the number of emitted photons in appropriate units, of a simple
candle with the brightness of these modern machines, using magnetic undulators for its amplification,
is presented in Fig. 8 . The graph shows that synchrotron radiation is about 10 orders of magnitude
brighter than the radiation of the sun, which emits the huge amount of energy of ~41 ergs per year. A
9
supernova, however, is radiating in much shorter times more than one order of magnitude more
energy. Figure 9 shows how this enormous amount of radiation energy is transferred into the kinetic
energy of high-speed acceleration of gas, expanding in all directions of the extreme-high-vacuum of
empty space.
For a more detailed study of vacuum physics, a list of a few general textbooks on high-vacuum
physics is given at the end of this article.
Fig. 8 Comparison of photon intensities emitted by various artificial light sources with the sun. (Courtesy Advanced Light
Source, Lawrence Berkeley National Laboratory.)
10
Fig. 9 A ring nebula observed by the Hubble Space Telescope (NASA/ESA, taken from CERN Courier, Vol.39, No.4, May
1999, p.10). The hot gas thrown into space by a supernova explosion travels at more than 100 km/s.
Table 3
Conversion factors for pressure in various systems of units [8].
Ultra-High (XHV)
According to this wide range of vacuum technology, vacuum measuring and vacuum generating
techniques have had to be developed for widely differing magnitudes, exploiting various physical
properties of gases. Table 5 shows the pressure ranges of the huge variety of vacuum gauges and
vacuum pumps, to be discussed in the following lectures.
12
Table 5
Pressure ranges of vacuum gauges and vacuum pumps.
-12 -10 -8 -6 -4 -2 0 2 4
Pa 10 10 10 10 10 10 10 10 10
U-tube
VACUUM GAUGES
Bourdon gauge
Diaphragm gauge
Capacitance
Thermistor
McLeod gauge
Penning gauge
Sorption pump
Cryopump
ULTRA-HIGH-VACUUM PUMPS
Diffusion pump
Turbomolecular pump
3. COMPOSITION OF AIR
Whereas atmospheric air is a mixture of gases with over 99% nitrogen and oxygen (see Table 6, for
instance) the rest gas of high and ultrahigh vacuum consists mainly of the lightest gas hydrogen. The
reason is that hydrogen easily penetrates walls, is adsorbed by all surfaces surrounding the volume and
is less effectively pumped than gases of larger active molecules.
Table 6
Composition of atmosphere (at sea level) [8].
p1V1/T1 = p2V2/T
. 2 = const
This means that for a given gas pV/T is a constant value. Under normal conditions this is best
fulfilled for H2 and He. The constant depends for a given temperature only on the number of molecules
contained in the volume, i.e. from the amount of gas enclosed.
For a given mass of a gas, three fundamental relations can be derived from Boyle-Mariottes
law, namely:
Boyles law Charles law Gay-Lussacs law
T = const p = const V = const
p1V1 = p2V2 = const V1/T1 = V2/T2 = const p1/T1 = p2/T2 = const
(isothermal phase transition) (isobar phase transition) (isochore phase transition).
14
For the case of thermal insulation one has Q = const , with dQ = 0 and the (p,V)-relation : pV =
const , corresponding to an adiabatic phase transition (dQ = dU + pdV with the specific heat capacities
at p,V = const).
It was observed that by relating pV/T to the concept of mole, its value is the same for any ideal
gas. A mole is the weight in grams equal numerically to the molecular weight of a substance. For
instance: 1 mole O2, H2, H2O = 32 g, 2.016 g, 18.016 g respectively. Avogadro found that the mass m
(of a standard volume of gas at the same temperature and pressure) is proportional to its molar mass
(molecular weight) M. In other words the amount of moles nM in a given mass m of a substance having
a molar mass M is: nM = m/M. It was shown by experiments that under standard conditions of
temperature (0 0C or 273.16 K) and pressure (1.013250 105 Pa or 760 Torr) one mole (or gram
molecular weight) of any gas occupies a volume of 22.4 liter (= 22 415 cm3). It is: pV = 22.4 m/M,
with m: mass of gas, M: molar mass, V in liters and p in Pa, where the pressure p is proportional to the
gas density = m/V.
For a gas mixture, the total pressure is the sum of all partial pressures: p = p1+p2+...+pi = pi
(Daltons law) and the mean molar mass is: <M> = (m1+m2+...+mi) / (nm1+nm2+...+nmi) (in g/mol).
The number density n of particles (atoms, molecules etc.) is given by:
n = N/V = (NA/R) (p/T) = nM (NA/V) = (m/M) (NA/V) (particles/m3),
where N is the total number of particles (molecules) and V the volume. Under the same conditions of
pressure and temperature, equal volumes of all gases have the same number of particles (Avogadros
law, L.R.A.C. Avogadro : 1776 - 1856). This volume is called a mole, and the number of particles in
one mol of a substance is given by Avogadros number NA. By independent measurements of the mass
of the hydrogen atom (particle mass) of mp = 1.67 10-24 g and the molecular weight of H2: M =
2.016 g, one can calculate Avogadros number to be :
NA = M/mp = 2.016 / (2 1.67 10-24) = F/e = 6.0221367 1023 (molecules/mol).
One gets the number NA independently also from the quotient of the value of the Faraday F
(defined as the electrical charge necessary to deposit electrolytically one mole of a substance) and the
charge e of an electron. One mole of any gas has NA particles and under standard conditions occupies
22.4 liters. From these values one can calculate that in a standard liter there are:
(6.022 1023 particles) / 22.4 liters = 2.69 1022 particles/liter .
R is the universal (or molar) gas constant:
R = 8.314 J mol-1 K-1 (with p in Pa, V in m3, T in K).
Whereas the universal gas constant R is related to moles, Boltzmann s constant k is calculated
on a molecular basis:
k = R/NA = 1.3806 10-16 erg/K = 1.3806 10-23 J/K , (with 1 J = 1.0 107 erg = 2.388 10-1 cal).
From this the equation of state of an ideal gas is derived, which describes how the measurable
quantities of a gas depend on each other:
pV = nM RT = (m/M) RT
Again, from Daltons law one gets for a mixture of gases: pVM = (nMi)RT, with total volume
VM = Vi and total pressure p = pi , if all components have the same pressure.
With n = N/V (molecules/m-3) and m = M/NA one obtains for the gas density: = mn = pM/RT =
pm/kT = M/VM. The density of gas depends on the two parameters temperature and pressure, which
both must be given, if the density is quoted.
The pressure p exerted by a gas depends only on the number density of molecules n = N/V and
temperature T of the gas, not on the particular gas. It can be expressed as the force F it exerts on a unit
area A or as the number density of molecules n = N/V, Boltzmanns constant k = R/NA and the
temperature T (see: section 2).
15
The correction term nM2 a/V2 ~ (m2/V2) = 2 with a = const takes the forces of intermolecular attraction
into account, which brings molecules closer together, acting like an additional pressure. The volume-
correction term nMb is a measure of the actual volume of all molecules of their own, which is four
times that of the molecule itself and increases with the molecular diameter. Whereas the isotherms of
an ideal gas are hyperbolas, those of a real gas at sufficiently low temperatures exhibit first a
maximum and then a minimum with decreasing volume (see Fig. 10). Normally, at critical values of
temperature Tc, volume Vc and pressure pc these extrema merge at a critical point into a curve with
horizontal turning tangent. The values of a and b are determined for most gases as a function of the
critical point (Tc, Vc, pc) of the corresponding gas (a = 3 pc (Vc /nM)2 ~ c-2 ; b = 3 Vc /nM) and are
tabulated. The limiting temperature TB (Boyles point) up to which a gas can still be described like an
ideal one, even up to high pressures, is given by TB = a/(bR).
Table 7
0
The root-mean-square velocity vrms of different gas molecules at 0 and 23 C [8].
In contrast to the velocity, the concept of temperature of a gas cannot easily be associated with a
single molecule. Temperature is understood in terms of the kinetic energy and velocity of motion. If
the air, surrounding us, has a molecular velocity below 400 m/s, it appears to us too cold, if it is above
500 m/s it is too hot.
5.2 Velocity distribution and mean free path
The constant motion of molecules and their collisions produce a wide distribution of velocities, from
about zero to very large values, which is expressed by the Maxwell-Boltzmann law (J.C. Maxwell:
1831 - 1879; L. Boltzmann: 1844 - 1906):
where dn is the number of particles in the velocity range between v and v + dv (per unit of velocity
range) and fv is the fractional number of particles between v and v + dv, which is zero for v = 0 and
v = and has its maximum at the most probable velocity: vp = (2kT/m)1/2 = (2RT/M)1/2 = 0.816 vrms. In
Fig. 11 fv is plotted versus v/vp
vp < va < vr ; v / vp = 1
0.8
0.6
fv
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3
v / vp
The average distance travelled by all gas molecules between successive collisions is called the
mean free path . It depends on the size and the number of the molecules, i.e. on gas density and
pressure. Applying the kinetic theory of gases by using the Maxwell-Boltzmann distribution and the
relation p = nkT, the average mean free path is given in m by:
= 1/(21/2 d2 n) = kT/(21/2 d2 p) = 3.107 10-24 T/(d2 p) ,
where d(m) is the molecular diameter, n = N/V (m-3) is the number of molecules per volume, T (K) is
the temperature and p (Pa) is the pressure. Accordingly, is proportional to the absolute temperature
and inversely proportional to the gas pressure.
17
For air (assuming that all molecules have the same d at room temperature (23 0C), one gets:
Air(m) = 6.5 10-3/p (Pa) . For common gases, the range of variation of is about a factor of 10.
When the mean free path becomes greater than the dimensions of the vacuum vessel, the collisions
become less frequent and the molecules mostly collide with the walls of the vessel.
The mean free path was derived for pure elastic collisions between the molecules. Taking into
account the attractive forces between the molecules, one found empirically that at very high
temperature T the mean free path is: T = / (1 + C/T), (with n, C = const).
It is interesting to compare the mean free path of electrons e (cm) and ions Ion (cm) in a gas
with that of neutral molecules (cm). One finds approximately: Ion = 21/2 = e/4. Values of mean
free path for different pressures or particle densities, respectively, are given below in Table 8.
Table 8
Rough values for characteristic vacuum parameters at a few pressures and gas flow regimes.
Pressure
5 2
10 Pa 10 Pa 10-1 Pa 10-4 Pa 10-7 Pa
103 mbar 100 mbar 10 mbar 10-6 mbar
-3
10-9 mbar
Particle Density, n (cm-3) 1019 1016 1013 1010 107
Mean Free Path, (cm) 10-5 10-2 10 104 107
(= 100 km)
Impingement Rate, ZA (s-1 cm-2) 1023 1020 1017 1014 1011
Collision Rate, ZV (s-1 cm-3) 1029 1023 1017 1011 105
Monolayer Time, 10 ns 10 s 10 ms 10 s 3h
Type of Gas Flow viscous Knudsen molecular
From the Maxwell-Boltzmann equation given above, one can derive the number of molecules
striking an element of surface perpendicular to their direction of movement, per unit time, the so-
called impingement rate ZA as:
ZA = 8.333 1022 p/(TM)1/2 (molecules m-2 s-1) , with p (Pa), T (K) and M (kg/mol),
= 2.635 1022 p/(TM)1/2 (molecules cm-2 s-1) , with p (mbar), T (K) and M (g/mol).
Similarly one can calculate the collision rate ZV (in molecules cm-3 s-1), which is the number of
collisions in a unit volume per unit time.
For ultra-high-vacuum systems there is another quantity of interest, namely the monolayer time
, which is defined as the time required for a monolayer to form on a clean, gas-free surface. To derive
, it is assumed that every impinging gas molecule finds and binds to a vacant site (sticking coefficient
1). The monolayer time , in seconds, is inversely proportional to the pressure and can be estimated
from:
= 3.2 10-6/p , with p in mbar.
Approximate values for particle density n , mean free path , impingement and collision rate ZA
and ZV and monolayer time are given for different pressures in Table 8, where also the relevant types
of gas flow are indicated (see also section below).
= 0.499 n m va ,
with the gas density = n m , va = average molecular velocity and = mean free path.
The specific heat conductivity of a gas is proportional to its coefficient of viscosity. Therefore,
they have the same dependence on pressure. Differences in particle density in a system lead to mass
transport, called diffusion.
The mean free path of gas particles is related to both, the thermal conductivity and the viscosity
of gases. Concerning heat conductivity and other phenomena of energy transport, one must distinguish
between low and high pressures and, correspondingly, between different states of a gas or types of gas
flow, which will be discussed in the following section. Depending on the state of the gas, transport
phenomena behave differently. Above approximately 1.3 kPa (= 13.3 mbar = 10 Torr) the heat transfer
through a gas inside a small chamber is dominated by convection due to the bulk motion of the gas. In
other words, in this so-called viscous state of a gas the totality of molecules is responsible for the heat
transfer. In this range of gas flow the heat conductivity of gases is independent of pressure, which is
also true for the viscosity. At lower pressures, when the mean free path of molecules becomes much
greater than the dimensions of the container, which is called molecular state, the individual molecules
carry the heat from wall to wall. Then the gas is no longer characterised by a viscosity. Under this
low-pressure condition of so-called free molecular conduction the heat transfer depends only on the
number of carriers. Thus the rate of energy transfer, the conductivity, is proportional to the pressure
and the temperature difference.
Both, the variation of conductivity and of viscosity with pressure are utilised for pressure
measurements, in the first case by monitoring the heat transfer rate from a hot filament to the
surrounding walls (thermal conductivity gauges), in the second case by measuring the deceleration of
a freely rotating body, a metallic sphere suspended in a permanent magnetic field, due to collisions
with the rest gas (viscosity manometers, spinning rotor gauge). Heat transfer in high-vacuum is very
low compared to conditions at atmospheric pressure. Therefore, there is very little heat flow for
instance between two metal plates in high vacuum. Heat transfer by conduction takes place only
through the small areas of contact, if not high clamping forces are used to enlarge the touching area.
7 GAS FLOW
velocities the flow is laminar, i.e. the layers are moving parallel, forming e.g. in a pipe a parabolic
velocity distribution with increasing velocities from the wall towards the axis of the pipe. Whereas the
limit between laminar, intermediate and molecular flow is defined by the value of the Knudsen
number, the limit between turbulent and laminar flow with the characteristics of a continuous fluid is
described by the value of another dimensionless quantity, the Reynolds number (O. Reynolds: 1842-
1912), which is expressed by
R = d v /
the ratio of kinetic energy and frictional work, with d = diameter of the pipe or characteristic size of
the vacuum component, v = gas velocity and , the density and viscosity of the gas at the
temperature of the flow.
By evaluating the Knudsen and Reynolds number, one can predict the various flow regimes,
which can be distinguished approximately by the following relations :
(1) R < 1200 ( K < 0.01 ) : laminar (viscous gas state)
(2) R > 2200 ( K < 0.01 ) : turbulent (viscous gas state)
(3) 1.0 > K > 0.01 : intermediate (transition gas state)
(4) K > 1.0 : molecular (rarefied gas state).
The exact value of R for which the flow changes from laminar to turbulent depends on the
geometry of the component, its surface roughness and other experimental factors. During evacuation
of a vessel, turbulent flow normally occurs only for a short period of time at the beginning. It has been
found to be approximately proportional to the root of the pressure gradient.
Transport phenomena in the viscous state of a gas are characterised by its coefficient of internal
friction or viscosity , which is defined as the tangential force per unit area for unit rate of decrease of
velocity with distance. The tangential force per unit area is given by the rate of momentum transfer
between adjacent layers. From this one can derive the coefficient of viscosity (see section 6):
= 0.499 va ,
in (units of viscosity) poise (J.-L.M. Poiseuille: 1799 - 1869), 1 Pa s = 1 kg m-1 s-1 = 10 poise .
Contrary to liquids, for which the viscosity decreases as the temperature increases, the viscosity
of gases at normal pressure increases with temperature. At very high pressures, however, the inter-
molecular forces are important and the momentum transfer is very different. Correspondingly, at very
low pressures there are almost no collisions between molecules, and momentum transfer happens only
between molecules and walls. Since the viscosity in the viscous state is not a function of pressure,
thermal conductivity is in this case also independent of pressure.
Like viscosity, diffusion of gases into each other is also determined by the mean free path . In
the case of diffusion of molecules in the same gas at constant temperature, the coefficient of self-
diffusion has been found to vary inversely as the density of gas or directly with : D11 = 1.342 /
= va /3. Diffusion pumps function on the principle of interdiffusion (coefficient D12) of a gas having a
low number density of molecules n = N/V into another gas that has a very high number density.
Transport phenomena in the molecular state, when is very large, are no longer determined by
molecular collisions or viscosity of the gas. In this case, molecules condense on a surface, rest on it a
given time and then reevaporate in arbitrary directions. There is a certain time required to form a
monolayer on a clean surface (see section 5.2 and Table 7). If the surface is in motion it can transfer a
velocity component to the molecule. This is the principle exploited by (turbo) molecular pumps.
Rotating molecular gauges (like the spinning rotor gauge) also use this principle of molecular drag.
Unlike the viscous state, in the molecular state heat conductivity or so-called free molecular
conduction depends upon pressure. In this range the rate of energy transferred by molecules between
two surfaces is proportional to the pressure and temperature difference between the surfaces. Thermal
conductivity at low pressures is used for measuring the pressure by thermal conductivity gauges.
Normally, with such gauges the pressure is determined by variation of temperature, keeping the energy
input for heating a filament constant.
20
If two chambers A and B with gases at different temperatures TA and TB are separated by a
narrow pipe or a porous plug, the following relation holds: pA TB1/2 = pB TA1/2. In this case, thermal
transpiration will occur, which means gas will flow between the two chambers until an equilibrium is
established.
This is important in situations where a chamber A is at low temperatures, say at liquid air
temperature TA = 90 K, and the pressure is measured with a gauge at room temperature TB = 300 K.
Then the real pressure is: PA = (90/300)1/2 PB = 0.55 PB. At higher pressures ( smaller than tube
diameter) molecular collisions become predominant and the condition of equilibrium is PA = PB
instead.
7.2 Conductance and gas flow for different vacuum components and systems
Q = p Sp = p dV/dt = Cp .
This equation holds only if p is constant, it does not apply to the transient state of flow. It
describes the throughput of a pump with pumping speed Sp at the intake pressure p, if p and V are
constant. Correspondingly, the throughput for a passive element can be expressed as the product of the
conductance C with p, where p is the difference between the pressures at the entrance and exit of
the element. By analogy with this expression, the pumping speed at any point of the vacuum system is:
S = Q/p .
This leads to the following expression for the effective pumping speed Seff of a pump with
nominal pumping speed Sp and the conductance of connections between the pump and the volume to
be evacuated :
1/Seff = 1/Sp + 1/Ctot
This is a basic formula in vacuum physics, because in most cases a pump is not connected
directly to a vacuum vessel. Almost always an intermediate system of pipes and various vacuum
components is necessary, which introduces a flow resistance (Ctot = total flow conductance) resulting
in an effective pumping speed Seff smaller than the pumping speed Sp of the pump. In order to ensure a
definite effective pumping speed at the container, a correspondingly higher speed of the pump must be
chosen. Only in the case where Ctot = (or impedance Ztot =0) is Sp = Seff.
From the equation given above one sees also, that it does not make sense to increase the pump,
if the conductance of the connecting components (pipes etc.) is limiting the pumping speed. One
should always make sure that in a certain pressure region only the corresponding relevant conductance
value is used. In general, it must be emphasised that the conductance of a vacuum component has by
no means a constant value, independent of the pressure. It depends greatly on the kind of flow, viscous
or molecular (see below), and therefore, on the pressure.
For air at 23 0C this gives: Qvis(air) = 4.97 (d4/L) pa (p1 - p2) , (Pa m3 h-1) .
The laminar conductance of straight pipes of uniform, circular cross section is :
22
Cvis = ( d4/128 L) pa (cm3 s-1) and for air at 23 0C: Cvis(air) = 4.97 (d4/L) pa (m3 h-1).
The conductance for an aperture of surface area A (in cm2) in the case of viscous laminar flow of
air is given by the formula of Prandtl (L. Prandtl: 1875 - 1953) :
Cvis = 76.6 0.712 (1 - 0.258)1/2 A/(1 - ) (l s-1) , with = p2/p1 < 1 .
At the critical pressure ratio (see above): crit = (p2/p1)cri t = 0.52 the conductance for an aperture
Cvis amounts to 40 A liters s-1 and approaches for 0 the limiting value Cvis = 20 A (l s-1).
Only a few examples for conductance have been given here. Formulae for conductance values
of vacuum components of different geometries and numerous nomograms for determining
conductances can be found in the literature.
The steady flow of gas in the transition regime is also called Knudsen flow. This happens when
the mean free path of molecules is of the same order of magnitude as the transverse dimensions of the
pipe through which the gas flows. Approximate formulae for this condition have been derived for
example by combining the laminar flow conductance with the molecular conductance of a long pipe.
If the vacuum system contains bends these can be considered on the basis that a greater effective
length is assumed, which can be estimated in the following way:
Ltot < Leff < Ltot + 2.66 n r ,
with Ltot, Leff : total and effective length (in cm), n the number of bends and r the tube radius.
Normally, most components are constructed such that the pumping speed is not significantly
throttled. As a general rule, one should always construct a vacuum line as short and wide as possible
and with at least the same cross-section as the inlet port of the pump.
The values for the laminar and molecular conductance of air, given above, should be multiplied
by the factors given in Table 9 for working with other gases.
Table 9
Conversion factors of conductance for air to other gases.
Table 10
Approximate outgassing rate K1 for several vacuum materials, after one hour in vacuum at room temperature.
-1 -2
Material K1 ( mbar l s cm )
Aluminium (fresh) 9 10-9
Aluminium (20 h at 100 0C) 5 10-14
Stainless steel (304) 2 10-8
Stainless steel (304, electropolished) 6 10-9
Stainless steel (304, mechanically polished) 2 10-9
Stainless steel (304, electropolished, 30 h at 250 0C) 4 10-12
Perbunan 5 10-6
Pyrex 1 10-8
Teflon 8 10-8
Viton A (fresh) 2 10-6
9. CONCLUSION
This introduction to a special branch of classical physics attempts to cover the fundamental principles
and most important phenomena of vacuum physics and technology described in much more detail in a
series of textbooks. It was written with the intention of providing a short summary and useful
background for the students of the CERN Accelerator School 1999 on Vacuum Technology, and to
help them better understand the subsequent lectures on current areas of research in vacuum physics.
BIBLIOGRAPHY
[1] P.A. Redhead, The Physical Basis of Ultra-High Vacua, Chapman & Hall, London (1968).
[2] G.L. Weissler and R.W. Carlson (editors), Vacuum Physics and Technology, Methods of
Experimental Physics, Vol. 14, Academic Press (1979).
[3] M. Wutz, H. Adam, and W. Walcher, Theorie und Praxis der Vakuumtechnik, F. Vieweg &
Sohn, Braunschweig/Wiesbaden (1986).
[4] H.J. Halama, J.C. Schuchmann and P.M. Stefan (editors), Vacuum Design of Advanced and
Compact Synchrotron Light Sources, Am. Inst. Phys., Am. Vacuum Soc. Series 5, Conf. Proc.
No. 171, BNL, Upton, New York (1988).
[5] A. Chambers, R.K. Fitch and B.S. Halliday, Basic Vacuum Technology, Adam Hilger, Bristol
(1989).
[6] A. Roth, Vacuum Technology, North-Holland, Elsevier Sci. Publ. B.V., 3rd updated ed. (1990).
[7] Y.G. Amer, S.D. Bader, A.R. Krauss and AR.C. Niemann (editors), Vacuum Design of
Synchrotron Light Sources, Am. Inst. Phys., Am. Vacuum Soc. Series 12, Conf. Proc. No. 236,
Argonne, Illinois (1990).
[8] A. Berman, Vacuum Engineering Calculations, Formulas, and Solved Exercises, Academic
Press, Inc. (1992).
[9] M. Prutton, Introduction to Surface Physics, Oxford Sci. Publications, Clarendon Press (1994).
[10] M.H. Hablanian, High-Vacuum Technology, a Practical Guide, Marcel Dekker, Inc. (1997).
[11] C. Edelmann, Vakuumphysik - Grundlagen, Vakuumerzeugung und -messung, Anwendungen,
Spektrum Akademischer Verlag (1998).